Omni-Directional Perception Based on Multiple Cameras: A Feasible Solution to the Limited Sensing of Wireless Multimedia Sensor Networks

In recent years, wireless multimedia sensor networks (WMSNs) attract great attention. Generally, WMSNs node is equipped with only one single camera for visual observation. In order to achieve wider view-angle, existing solutions, such as node rotation and view-angle adjustment, always cause high cost, high energy consumption and non-real-time response. In this paper, we propose a feasible solution by utilizing four wide-angle cameras to form omni-directional visual sensing for WMSNs node. The omni-directional probability sensing model is analyzed and defined. With DSP processor, we carry out the system architecture in details. Energy consumption is analyzed by comparing with rotatable camera. To our best knowledge, it is the first time that omni-directional vision with multiple cameras is introduced for WMSNs node to monitor the surrounding environment in real time. Experiment results show that the proposed omni-directional sensing can achieve better performance on deployment and energy cost than traditional WMSNs node.